The gap between game-theoretic predictions and actual choices people make in, for instance, gaming experiments has been over-interpreted as evidence against rationality of players. I consider a version of the ultimatum game and examine its equilibria under different assumptions about players’ preferences. Using standard notions of rationality I show that the discrepancy between the “normative” and the “descriptive” cannot be established by a simple comparison of what is predicted by the equilibrium choices and the actual choices